From product recommendations and personalised offers to AI shopping assistants and customer service bots, artificial intelligence is becoming part of the buying journey. While recent research points to faster product discovery and greater convenience, it also reveals concerns around inaccurate recommendations, disconnected customer data and the difficulty of reaching human support.
Imagine shopping for a new laptop without opening multiple websites, comparing specifications or spending hours reading reviews.
Instead, a customer asks an AI assistant to recommend three models within a particular budget. The assistant compares battery life, performance and prices before narrowing the options.
The customer makes a purchase and receives an automated confirmation. Everything appears straightforward.
A few days later, however, the delivery is delayed. The retailer’s chatbot repeatedly offers tracking information but cannot explain what happened. When the customer requests human assistance, the conversation begins again.
What started as a convenient shopping experience has become frustrating.
This contrast is becoming increasingly relevant as artificial intelligence moves deeper into marketing technology.
AI is no longer limited to helping brands create advertisements or analyse campaigns. It is influencing how consumers discover products, compare alternatives, receive recommendations and interact with businesses after making a purchase.
For marketers, these capabilities offer opportunities to improve personalisation, reduce repetitive interactions and respond to customers more quickly.
But whether they improve the overall buyer experience is less certain.
Salesforce’s 2026 State of Marketing findings for India, based on 250 marketing decision-makers within a global survey of 4,450 respondents, found that 81% of Indian marketers had adopted AI.
The study also found that 92% believed customers increasingly expected two-way conversations with brands, while 86% said they would trust AI to help respond to customers at scale.
Yet 71% struggled to respond promptly because they could not access the customer context they needed.
The findings highlight a central challenge for AI-driven marketing. Technology can accelerate communication, but a faster response does not necessarily mean a better one.
Recent research from Deloitte, Salesforce, Qualtrics, Capgemini and Adobe suggests that the impact of AI depends on which parts of the journey are automated and how effectively those systems understand the buyer.
Is AI Becoming the First Place Buyers Look?
Product discovery is among the areas where AI is changing established consumer behaviour.
Traditionally, shoppers searching for a product might visit several websites, read reviews, compare prices and consult friends before making a decision.
AI assistants can bring parts of that process into a single conversation.
A customer looking for a refrigerator could specify a budget, preferred capacity and energy-efficiency requirements. Instead of independently comparing dozens of listings, the customer could ask an assistant to identify suitable options and explain their differences.
Deloitte’s 2026 research involving 13,500 consumers across 15 European countries found that 56% had used AI for shopping at least once.
Among those using AI for shopping, 57% used it to compare products. The research indicated that AI was becoming an important comparison channel, although traditional retailers and marketplaces continued to play significant roles when consumers completed purchases.
This distinction is important for brands.
Consumers may increasingly arrive at a retailer’s website after an AI assistant has already influenced their shortlist.
That changes how businesses compete for attention.
Product information may need to be sufficiently clear, accurate and accessible for AI systems to interpret it alongside competing options.
However, a recommendation is only as reliable as the information supporting it.
An assistant comparing two smartphones might overlook a recent price change, rely on outdated specifications or present an incorrect claim about availability.
The shopper may save time during research but still need to verify the recommendation before purchasing.
Deloitte’s Morris Boermann, Director and Business Lead, observed that “Trust is earned through experience, not messaging.”
The statement reflects an important distinction between influencing a buying decision and delivering a satisfactory purchase.
AI can introduce customers to products they might otherwise overlook. But its usefulness ultimately depends on whether those recommendations are accurate and relevant.
For marketers, this also raises questions about visibility. As more consumers use AI to compare alternatives, brands may need to consider how their products are represented in AI-generated answers, not just conventional search listings.
When Does Personalisation Become a Problem?
Personalisation has long been central to marketing technology.
Brands use browsing behaviour, purchase histories and customer preferences to deliver more relevant messages.
AI can make those processes faster and more detailed, generating recommendations or campaign variations for different customers without requiring every message to be created manually.
However, personalised communication can become ineffective when customer data are incomplete.
Salesforce’s India findings revealed that 83% of marketers needed more personalised content than they could currently produce, while 81% were using AI to help address that demand.
At the same time, 98% reported barriers to personalisation, including privacy concerns, poor data quality and technical limitations.
Only 60% reported complete access to service data, while 61% had complete access to sales data and 58% to commerce data.
Consider a customer who recently booked an international holiday through a travel company’s call centre.
An automated marketing system may continue recommending the same destination because it cannot access the completed booking.
The messages may match the customer’s previous interest, but they no longer reflect the customer’s situation.
A similar problem can occur when a retailer sends repeated discounts for a product someone has already purchased.
These are examples of how AI can magnify existing gaps in marketing systems.
“The biggest barrier to personalisation today isn’t AI, it’s the quality and connectedness of the data that powers it,” said Nishant Kalra, Vice President, Sales, Salesforce South Asia.
The challenge extends beyond connecting customer records.
A promotion can be relevant to someone’s interests while still arriving at an inappropriate moment.
For instance, a customer waiting for a refund may be less receptive to a new sales offer, regardless of how accurately the recommendation reflects previous purchases.
AI can identify patterns and generate messages, but those capabilities do not automatically account for the broader relationship between a customer and a business.
This creates an important difference between delivering more personalised communication and improving the buyer experience.
The latter depends on whether the message is useful, appropriately timed and consistent with what the customer has already experienced.
Why Can a Friendly Chatbot Still Leave Customers Frustrated?
Customer service offers one of the clearest examples of AI’s benefits and limitations.
Automated assistants can handle routine requests, provide information outside business hours and reduce waiting times.
For straightforward questions about delivery status, opening hours or product availability, these capabilities can make the experience more convenient.
Problems become more apparent when a customer requires assistance beyond a predefined process.
Qualtrics’ 2026 Agent Effectiveness Benchmark research, involving 7,001 consumers across seven countries, found that AI-handled interactions scored 4.41 for friendliness but 4.08 for understanding.
Understanding recorded the weakest result among the evaluated qualities of AI-assisted interactions.
The study also found that overall customer satisfaction fell to 2.83 when issues remained unresolved. Consumers rated an agent’s understanding 37% lower in unresolved interactions than in resolved ones.
“Your AI agent might be polite, but our research shows it lacks the context to turn the experience into a meaningful outcome for the customer,” said Leonie Brown, Head of Qualtrics Labs.
The findings highlight why conversational fluency should not be confused with problem-solving ability.
A chatbot may apologise politely and provide a reassuring response. But if it cannot recognise an earlier complaint or escalate an unusual issue, the customer may still leave dissatisfied.
Capgemini Research Institute’s 2026 customer experience research, covering 9,500 consumers across 16 countries, found that 66% ranked front-line employees among their three preferred interaction channels.
By comparison, 56% included generative AI-enabled bots or agents among their top three.
Another 36% said they would be willing to pay a premium for support from front-line employees.
These findings do not imply that customers reject automated assistance. Rather, they indicate that human interaction remains valuable, particularly when decisions become complex or emotionally significant.
A customer may accept AI assistance while checking a delivery date but prefer a person when seeking an exception to a cancellation policy.
The distinction also matters in sectors such as banking, insurance and travel, where unresolved issues can have considerable financial or personal consequences.
“For personal financial decisions, customers place greater trust in human professionals than AI-driven bots,” said Hitender Sachdeva, Head of Merchant Service, Products and Payments at Scotiabank, in Capgemini’s report.
For marketing and customer experience teams, the question is therefore not simply how many conversations automation can handle.
It is whether customers receive appropriate assistance when automated systems reach their limits.
Do Buyers Want AI to Take Over the Entire Journey?
The next phase of AI-powered commerce could involve systems capable of completing more tasks on behalf of consumers.
Rather than only suggesting products, AI agents could increasingly assist with selecting options, managing bookings and carrying out purchasing-related activities.
However, interest in these capabilities does not necessarily mean consumers want to surrender control.
Adobe’s 2026 AI and Digital Trends consumer research, based on 4,000 customers across global markets including India, found that 43% were willing to try a brand’s AI concierge or agent.
Yet only 19% wanted AI agents to become their primary method of interacting with brands.
The study also found that 37% would stop interacting with a brand if they discovered they were communicating with AI when they had expected a person.
Respondents identified the ability to switch to human assistance at any time as the most important safeguard.
The findings suggest that consumers distinguish between using AI for convenience and allowing it to manage the complete relationship.
For example, someone researching health insurance may appreciate an AI-generated comparison of premiums and benefits.
The same customer might prefer to speak to an adviser before selecting a policy involving exclusions, long-term financial commitments or complicated conditions.
Similarly, an AI assistant could help identify suitable holiday packages without necessarily being authorised to complete a costly booking.
The preferred level of automation may vary according to the financial risk, complexity and personal importance of the decision.
It may also vary between consumers.
Deloitte’s European research found differences in attitudes between generations. While 21% of Gen Z respondents preferred asking an AI assistant for help while shopping in a physical store, 48% of Gen X respondents preferred human assistance in that situation.
These differences make it difficult for brands to assume that one automated journey will satisfy every customer.
They also raise questions about how businesses measure the impact of AI.
Faster response times, lower service costs and more automated conversations are useful operational indicators, but they do not independently demonstrate improved buyer satisfaction.
A recommendation that leads to a purchase may appear successful until the product fails to match the customer’s requirements.
Similarly, a chatbot that closes an enquiry without employee involvement may still have left the underlying issue unresolved.
Businesses may therefore need to examine outcomes such as successful problem resolution, repeat enquiries, customer satisfaction and the ability to access alternative support channels.
Across the five studies, the evidence presents a mixed picture.
AI is helping consumers compare products, navigate information and receive quicker assistance. At the same time, incomplete customer records, inaccurate recommendations and unresolved service conversations remain sources of frustration.
The findings also suggest that customers are interested in using AI selectively rather than making automation the default for every interaction.
For marketers, the challenge is to determine where technology genuinely removes effort and where it introduces additional steps.
The future buyer experience may depend less on how much AI a brand deploys and more on whether customers feel that the technology understands their needs, resolves their problems and allows them to remain in control.
Disclaimer: All data points and statistics are attributed to published research studies and verified market research. All quotes are either sourced directly or attributed to public statements.